The Way Forward

Ideas
1

Summary: a natural reweight will not happen, and a major overhaul of the protocol is necessary. However, there is no need to panic.

I know many of you are worried that your FEI (and therefore TRIBE) will become digital trash. I am not worried at all. In fact, after I finish writing this, I am going to buy more FEI. This post will describe why I know FEI is extremely unlikely to fail, and what we need to do to make sure it never does. Let me start with a clear explanation of what is happening right now.

First, a clarification. There is an idea floating around that creating use cases of FEI will support its price. This is a misunderstanding. The utility of FEI comes from the (yet to be established) fact that it is pegged at $1, just like any other stablecoin. If FEI is always equal to $1, anyone will be willing to sell his coins, products, or services in return for FEI. If the price of FEI is not stable, it will never be demanded.

So the question, and the only relevant question, is this: “can FEI be pegged?”

Though this may surprise some people, the answer is a definite yes. As Joey as mentioned repeatedly, the fundamental backstop is the fact that there is $1,265,880,525 of Ethereum in the PCV, whereas there are less than 1,150,000,000 FEI held by users. We are overcollateralized. Forget about the fact that the ETH is called PCV, or that it’s in a Uniswap pool. Just think of it as gold held by a bank that backs up paper bills. The worst thing that can happen is no one wants to use FEI after all, and we decide to return the ETH to FEI holders, pro rata. That means each FEI will be worth 1,265,880,525/1,150,000,000=1.1 dollars. If you sell FEI now, you receive around 70 cents. For this to be worth it, you need to believe that price of ETH will fall to around $1300 by the time FEI fails. I am not betting my money on that.

So if the backstop is so strong, why is FEI falling? That is because the reweighting protocol is flawed. People are selling FEI at 70 cents, and yet a reweight is still more than 7 hours away according to Fei Protocol. In fact, as was pointed out by @Len in discord (and perhaps others that I am not aware of), under the current protocol, a reweight is unlikely to happen at all. The protocol is designed in such a way that every time someone purchases FEI and ticks the price of FEI higher, the time until reweight is increased by the same ratio. To use a bit of math, reweight happens when enough time passes such that w(t)=100m, where t is time, w(t) is linearly increasing in t, and m is the price distance from the peg. But whenever someone buys FEI and reduces m, the protocol decreases t and thus w(t). So as long as there is enough fluctuation of FEI price, w(t) can never reach the level required for reweight to happen.

This is not how it should be. We need to implement a new protocol that decides when a reweight happens. There have been multiple suggestions, summed up in the main thread. However, this is not something to be decided on the spur of the moment. It won’t be enough to simply force a reweight a few times. We must design a new protocol that can implement automatic reweights in order to truly peg FEI. This will not be easy, of course. However, because we are overcollateralized, we do have some time to carefully design a new protocol. To use this precious time wisely, let us start by laying out some of the high-level properties that the protocol should satisfy.

  1. Reweight must happen whenever a low enough price has been sustained for a long enough time.
  2. Protocol must not be gameable. E.g. it must not be possible for a user to purposefully induce or delay reweights and benefit themselves or hurt the protocol.
  3. Mint and Burn together must be deflationary.
  4. PCV value must be non-decreasing in activities.

The current protocol fails 1, as we are currently observing.

Below are some other properties that seem desirable, though perhaps not as essential as the above three.

A. Reweight must be more likely to happen when the price is lower, or when a low price has been sustained for a longer period of time.
B. Mint and Burn should provide incentives to support the peg, so that reweight does not happen too frequently.

Clearly, this is not an exhaustive list. There are probably other properties that the current protocol is designed to satisfy, but that I am not aware of. Our way forward now is to carefully write down all the properties that we need, and design a new protocol that satisfies these properties. It may not be possible to fully attain everything; some may have to hold probabilistically, others may hold as long as we are in some natural notion of equilibrium. This will not be an easy task, but it must be done for FEI to survive.

My message to the core team:
Acknowledge that the protocol has a flaw. Promise that, if all efforts fail, you will put to vote a proposal that entitles each holder of FEI to a pro rata share of the ETH in PCV. Letting this promise serve as a backstop, ask FEI holders to give ourselves some time to figure things out. Bring in the best minds to design a new reweight protocol. Announce this protocol, let everyone vet it, then put it to a vote.

My message to FEI holders:
The launch was far from perfect. However, it succeeded in one important aspect - we are overcollateralized. This buys us time. Let us allow the developers to use this time to do what they do best: design a protocol.

Now I’m off to buy some FEI.

16 Likes
2

The Tribe token gives the vote.

3

It buys time to fix protocol. But the more time that passes the more the people who supported the project feels burned. This might lead to difficulties using it in the future.

Who knows? I pre-swapped for TRIBE - difficult to stay positive after 30%+ burn. Expected a dip, but nothing like this

1 Like
4

it’s not clear from the white paper how the w(t) changes with purchase, it only vaguely mentioned how a purchase prolongs wait time. The only way for a reweight to happen, is for the burn rate and reward rate to be equal. The reward rate slowly increases linearly, however, it decreases when people buy FEI. At a time when reweight is very close, If NO ONE BUYS, the linear increment in reward rate would end up equating it to burn rate, triggering the reweight. However, it’s very difficult for anyone to NOT BUY when reweight is very close. In fact, people want to buy right before reweight so that they can immediately get a profit. But that very buying action moves reweight into the future. So this is ending up to be an endless delaying of the reweight.

I discussed this very point of this @Joey, this reduction of reward rate if someone purchases is a design choice. If there is no reduction in reward rate when purchasing FEI, people with large funds can intentionally trigger a reweight by bulk buying and moving up the FEI price (this will drop the burn rate to meet the reward rate), and then they can call the reweight function themselves, and then front run everyone else by selling right after reweight, creating a new negative peg again.

I feel like overall FEI is well designed, and the first two days is not working out so well, but there is no need to throw everything out just yet. From reading the white paper and discussing the various aspects in the discord and forum, it’s clear that the FEI team had put a lot of time and thought in designing the protocol. Inventing a new protocol ad-hoc will likely not work out so well, considering the various nuances that need to be considered, and it’s very likely that someone can find a way to game (or even worse, malicious hack) a system that’s not carefully thought through and planned out.

Some level of tweaking in FEI can be easily done and will probably be done to improve the protocol. In fact, in the other thread there are currently 5 short-term solutions proposed that will likely restore the peg. (I think all 5 are plausible options if done right). Although none of them address the problem of a (seemingly) infinite delay of reweighing.

An easy solution is avoid further delaying of the reweight is to just to have a time condition to trigger reweight: if FEI spends over x hours below 0.99, then the reweight function can be called. That dosen’t really need a new protocol, and it’s likely to work well to solve the current conundrum.

3 Likes
5

I agree that having a hard time x price condition is a good idea. I disagree that it is an obvious solution that can be implemented as a quick fix, or that it is obviously the best option we can come up with. Just as reweight can be endlessly delayed under the current protocol, it could be that people buy at time x-e for e>0 small and prevent the reweight. Does this mean we could get stuck at the threshold price of 0.99?

That said, I think this idea is much better than most of the 5 options in the other thread. For example, forcing a manual reweight a couple of times, or simply increasing the growth rate of the mint incentive do nothing to solve the flaw in the current protocol.

I think we agree that the current protocol is flawed. I don’t think this is something that can be fixed in a few days by having discussions on an anonymous board. Governance is meant to vote on well-thought out proposals, or natural changes in parameters. But first we need experts to develop and propose a proposal. Just as the team proposed a well thought out protocol before the launch. And fixing a protocol actually means redesigning it when the problem is to design a complex protocol that satisfies a number of design objectives.

By the way, https://docs.fei.money/ has some of the descriptions missing in the whitepaper, though I don’t think it’s crystal clear either.

6

For now! Crypto tend to lose 95% of its value for a short period of time. What will you say then?

1 Like
7

@jony @mochak Yeah, we can’t afford to take forever. But I think being careful and thorough about it is the only way to fix this.

8

If it comes to that, instead of having to reinvent a brand new stable coin on the fly, I’d rather FEI just to create a contract to let anyone redeem their FEI for ETH at 1:1 Peg, and be done with it.

2 Likes
9

This post says Fei is overcollateralized.
The Fei protocol site does not give a supply statistic,
but CoinGecko says the supply is over 2B FEI
Which would mean it is very under collateralized.
Why does CoinGecko show there is so much?

10

The circulating supply on CoinGecko or Coinmarketcap includes 1.34 billion FEI that is in the Uniswap pool. Most of these are held by the PCV, which means they are not really circulating.